Stratir

Stratir research brief 06

Frontier Health Systems

Clinical workstation in a community health setting

Research brief
Madagascar, Kenya, Eritrea

Public health and software systems

A source-backed research brief on why sparse and underconnected regions need different software architecture, and how Stratir builds products that run where specialist coverage and connectivity do not.

Abstract

Problem framing

Health systems in Madagascar, Kenya, and Eritrea operate under different governance models, but share a structural constraint: clinical decisions, stock movement, and surveillance signals must often be recorded far from reliable connectivity and specialist backup. National averages hide county, island, and highland variance that breaks cloud-only product assumptions.

This brief synthesizes published workforce, mortality, expenditure, and connectivity indicators to argue for offline-first, ontology-backed health software. The objective is not to replace ministries or NGOs. It is to show where a product studio can ship inspectable systems that reduce data loss, referral delay, and stock-out blind spots in the most underserved catchment areas.

Hypothesis

Software leverage is highest where physician density is lowest and mobile access is uneven.

Under-five mortality rate (deaths per 1,000 live births), modeled trend from WHO/UNICEF benchmarks
0255075100200020052010201520202023MadagascarKenyaEritrea

Trend lines approximate published WHO/UNICEF series for comparative framing. Latest anchor points match GHO country profiles used in the tables below.

Connectivity proxy vs physician density (conceptual gap analysis)
Mobile broadband adoption proxy (index 0-100)Physicians per 10,00000.511.52Software leverage zoneMDGKENERI

Kenya shows higher digital readiness but persistent rural access gaps. Madagascar and Eritrea sit in the quadrant where offline software, CHW tooling, and sync discipline matter most.

Physician density by country (per 10,000 population)
Madagascar0.21Kenya1.6Eritrea0.76

Country profiles

Three contexts, one design problem

CountryPopulationRural sharePhysicians / 10kU5MRHealth spend (% GDP)
Madagascar30.3M63%0.2151.23.8
Kenya55.2M70%1.635.54.9
Eritrea3.6M59%0.7641.84.2

MDG

Madagascar

Burden: Malaria, maternal mortality, tuberculosis, and malnutrition remain primary outpatient and inpatient drivers.

Connectivity: Low national broadband penetration; mobile-first access in coastal and urban corridors.

Constraint: Island geography, seasonal road disruption, and thin specialist coverage outside Antananarivo.

KEN

Kenya

Burden: HIV/TB co-infection, NCD growth in urban centers, and county-level referral bottlenecks.

Connectivity: Stronger mobile money and API ecosystem; uneven last-mile connectivity in arid north and remote counties.

Constraint: Devolved health system creates data fragmentation across 47 counties without shared ontology.

ERI

Eritrea

Burden: Maternal health, vaccine cold chain, and infectious disease surveillance across dispersed highland communities.

Connectivity: Limited public internet infrastructure; facility reporting often manual and batch-oriented.

Constraint: Sparse specialist network and high cost of real-time sync across remote clinics.

Methods

Research questions

RQ1

Which clinical and logistics signals can be captured offline at community level without breaking auditability?

RQ2

How do referral pathways differ across Madagascar, Kenya, and Eritrea when transport, connectivity, and county boundaries change?

RQ3

What minimum viable ontology lets a national program, NGO partner, and district clinic share the same patient and stock vocabulary?

RQ4

Where can bounded agentic automation reduce clerical load for nurses and CHWs while keeping human review on high-risk decisions?

System design

Product layers that match frontier health operations

Community capture

Offline-first mobile forms for vitals, symptoms, stock counts, and referrals with conflict-safe sync when connectivity returns.

Facility registry

Canonical facility, staff role, and service capability model so dashboards do not double-count mobile clinics or seasonal camps.

Supply and cold chain

Temperature excursion alerts, batch traceability, and reorder logic tuned for irregular delivery schedules.

Surveillance layer

Case aggregation with geospatial clustering for malaria, TB, and outbreak signals without exposing identifiable records in open views.

Review queue

Human-in-the-loop triage for abnormal vitals, stock-outs, and duplicate registrations before escalation to district teams.

Partner export

DHIS2-aligned or CSV export paths for ministries and donors that require standardized reporting windows.

Stratir delivery

Building software for regions the default stack ignores

Stratir applies the same studio discipline used for intelligence and agentic products to public health and humanitarian contexts: ontology first, bounded automation, human review on high-risk paths, and exports partners can audit. The goal is production software that survives sparse connectivity, shared devices, and multi-org reporting windows.

Discuss a health systems build

01

Epidemiologic scoping

Stratir maps the decision owners, indicator definitions, facility types, and reporting cadence before any interface is designed.

02

Ontology for sparse networks

Entities (patient, encounter, stock lot, referral, CHW route) are modeled once so county, island, and national views stay aligned.

03

Offline-capable product

Engineering prioritizes low-bandwidth sync, durable local storage, and role-based access on shared devices.

04

Pilot with measurable endpoints

Pilots track time-to-referral, stock-out duration, data completeness, and supervisor review turnaround, not vanity usage metrics.

References

Sources and indicators

WHO Global Health Observatory

Physician density, under-five mortality, and health workforce indicators.

World Bank World Development Indicators

Health expenditure (% of GDP), rural population share, and connectivity proxies.

UNICEF Data

Child mortality and immunization coverage benchmarks for cross-country comparison.

Africa CDC / ministry reporting frameworks

Regional surveillance and export conventions for partner alignment.

Indicator values in tables and charts are anchored to publicly reported WHO GHO and World Bank WDI country profiles for comparative research framing. They should be validated against the latest ministry and partner datasets before operational or procurement use. This brief is a systems research artifact, not clinical guidance.